Abstract
Brain decoding with functional magnetic resonance imaging (fMRI) requires analysis of complex, multivariate data. Multivoxel pattern analysis (MVPA) has been widely used in recent years. MVPA treats the activation of multiple voxels from fMRI data as a pattern and decodes brain states using pattern classification methods. Feature selection is a critical procedure of MVPA because it decides which features will be included in the classification analysis of fMRI data, thereby improving the performance of the classifier. Features can be selected by limiting the analysis to specific anatomical regions or by computing univariate (voxel-wise) or multivariate statistics. However, these methods either discard some informative features or select features with redundant information. This paper introduces the principal feature analysis as a novel multivariate feature selection method for fMRI data processing. This multivariate approach aims to remove features with redundant information, thereby selecting fewer features, while retaining the most information.
Highlights
Over the past decade, multivoxel pattern analysis (MVPA) has become widely used in functional magnetic resonance imaging because of its effectiveness in decoding cognitive states [1,2,3,4,5,6,7]
The results show that principal feature analysis (PFA) is an effective feature selection method for functional magnetic resonance imaging (fMRI) data processing because it can retain most of the information with fewer features
We used the voxels as features to study the performance of different feature selection methods (PFA, t-statistics, and searchlight)
Summary
Multivoxel pattern analysis (MVPA) has become widely used in functional magnetic resonance imaging (fMRI) because of its effectiveness in decoding cognitive states [1,2,3,4,5,6,7]. Unlike univariate statistical methods focusing on characterizing the relationship between cognitive variables and individual brain voxels, MVPA applies powerful pattern classification algorithms to multivoxel patterns of activity to decode the information that is represented in that pattern of activity [1]. A cross-validation procedure is required to accurately estimate the performance of the pattern classifier. This procedure is implemented by dividing the data of all samples into portions of equal size. Where A is the eigenvector of the covariance matrix Σ, written as A = Features that are highly correlated or have high mutual information will have similar weight vectors ai, which can be used to remove features with redundant information. PFA finds the highly correlated features and removes those with redundant information
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
